Summary

Unmesh Joshi, writing on Martin Fowler's site, argues that code serves two purposes: instructions for machines and a conceptual model of the problem domain. As LLMs commoditize the mechanical aspect of code generation, the conceptual modeling role becomes more valuable. The post draws heavily on Domain-Driven Design concepts like bounded contexts and ubiquitous language to explain how coding is fundamentally an act of vocabulary building through iterative collaboration. Well-structured code with precise vocabulary actually makes LLMs more effective, while LLM-generated code without shared understanding creates 'cognitive debt.' The conclusion is that coding's future lies not in producing syntax faster but in building better conceptual models.

Key Insight

As LLMs commoditize code generation, the enduring value of coding shifts from producing machine instructions to building precise conceptual vocabularies that both humans and LLMs can reason with.

Spicy Quotes (click to share)

• 5

  If producing code becomes cheaper, what remains valuable about code?

• 3

  A well-designed codebase does not only contain instructions for the machine; it also contains concepts for humans and tools to reason with.

• 3

  Coding for a domain is fundamentally an act of translation.

• 3

  The right abstraction often is not obvious upfront; it reveals itself only as you continually mold and refactor the code against real-world constraints.

• 4

  Frameworks and libraries are codified vocabularies.

• 5

  Cognitive debt accumulates when words, abstractions, and structures are used without their meaning being well understood by the people working with them.

• 7

  The problem is not that the LLM generated code. The problem is that the code introduced vocabulary faster than the developers built understanding.

• 4

  The future of coding is not just writing more code faster. It is building better conceptual models, better vocabularies, and better foundations on top of which both humans and LLMs can work.

Tone

reflective